The Moon landing is now a relay race: Artemis III tests both landers in orbit in 2027, Artemis IV lands in 2028

NASA has turned its return to the Moon into a relay, and on Friday it published the running order. Artemis III, still on track for summer 2027, will not land anywhere. It is a 14 day mission in Earth orbit where four astronauts in the Orion capsule will rendezvous and dock with test versions of both commercial lunar landers, SpaceX's Starship and Blue Origin's Blue Moon Mark 2, trying out in the safety of home orbit every manoeuvre the landing will later demand far from home. Artemis IV then attempts the first lunar surface landing of the programme in 2028.

The choreography is genuinely unlike anything flown before. Blue Origin launches its Blue Moon test lander first, on a New Glenn rocket, and it can wait in orbit for up to a month. NASA then launches the crew on the Space Launch System, Orion docks with Blue Moon on day three, and the astronauts climb inside to test its life support and controls. Then SpaceX launches a Starship test lander, Orion undocks from one lander and docks with the other for about two hours on day eight, a docking only test with no crew transfer. Artemis programme manager Jeremy Parsons called it likely the most complicated launch campaign ever undertaken, and for once the superlative is doing no decorative work: three of the most powerful rockets ever built, flying in sequence, on one mission clock.

Why rehearse at all, when the programme's critics want speed? Because the architecture changed underneath Artemis. NASA restructured the programme earlier this year after technical setbacks, standardising the SLS configuration and aiming to fly it roughly every 10 months instead of once every three years, and it has not yet chosen which company's lander will carry the Artemis IV crew down. That choice now depends on evidence this mission creates: uncrewed landing demonstrations through 2027, and then the crewed orbital tests, with each company effectively auditioning in front of the customer. Administrator Jared Isaacman framed Friday's briefing as acceleration, and the honest reading is that it is acceleration by reordering: the landing slips from the third mission to the fourth so that the third can retire the risks that would otherwise be retired at the Moon.

The race framing is not invented by headline writers. NASA officials said openly that the aim is to return American astronauts to the lunar surface before China and to build toward a permanent presence, and the schedule now holds two landings in 2028 if Artemis V follows later that year. A relay with two runners from rival companies is more resilient than a single bet, and more expensive, and more complicated, all at the same time. That is the trade NASA has chosen, in public, with dates attached.

Our call is that Artemis III is now the mission the whole programme will be judged by, precisely because it lands nowhere. A flawless 2027 rehearsal, two dockings, two landers, three rockets, makes the 2028 landing look engineered rather than hoped for, and gives NASA the evidence to pick a lander without picking a favourite. One failed docking, one lander not ready at the pad, and the relay drops the baton in full view, with the 2028 date dissolving into the familiar Artemis fog. The Moon can wait a year. The choreography cannot afford to fail even once, because there is no second rehearsal in the plan.